VERIFICATION OF FLOW ROUGHNESS COEFFICIENTS IN SMALL ALASKA STREAMS

Accurate estimation of the flow roughness coefficient, often expressed as Manning's n, is critical in surface-water models used for predicting water surface elevations or stream discharge. Flow roughness generally cannot be measured directly, and common methods for estimating Manning's n based on data from other streams may not be applicable to Alaska streams with steep gradients and coarse bed material. Manning's n for 12 small Alaska streams in various physiographic settings was verified from field measurements of discharge, water surface elevation, and two to four channel cross-sections. Width-to-depth ratios at high flow ranged from 10 to 40, drainage areas from 1.5 to 140 square miles, and water surface slopes from 0.4 to 7 percent. Gravel and cobble bed material provided grain roughness. Large boulders, poorly organized boulder steps, bedrock protrusions, mild contractions and expansions, and riffle-pool sequences provided form and spill roughness. Measurements at both flood conditions and at moderate to low discharge quantified the corresponding change in flow roughness.

Preliminary results on 10 streams suggest that field estimates by experienced hydrologists underpredict Manning's n for this type of stream. A common equation using friction slope and hydraulic radius overpredicted Manning's n for these streams. Spatial variability of computed n values within streams was apparent in computations for various combinations of two-cross-section reaches within streams. The variation was weakly related to the magnitude of n, such that two-cross-section n values varied from the all-cross-sections n by an average of plus or minus 15 percent of the all-cross-sections n for streams with n in the range 0.032 to 0.096. Combined with observations of roughness-generating features, this suggests that spatial variability is inherent to all the streams but particularly prevalent in streams with high values for form and spill roughness. All-cross-sections n values were correlated to slope and inversely correlated to hydraulic radius, suggesting that these variables may provide a statistically valid regression equation for estimating flow roughness coefficients in similar streams and improving the accuracy of surface-water models for this stream type.